Article(id=1239158368489754638, tenantId=1146029695717560320, journalId=1238841944844054536, issueId=1239158367571210992, articleNumber=null, orderNo=null, doi=10.12347/j.ycyk.20250723001, pmid=null, cstr=32406.14.ycyk.20250723001, oa=null, hot=null, price=null, onlineType=0, articleFormat=0, articleType=null, articleTypeStr=null, receivedDate=1753200000000, receivedDateStr=2025-07-23, revisedDate=1753718400000, revisedDateStr=2025-07-29, acceptedDate=null, acceptedDateStr=null, onlineDate=1773367977496, onlineDateStr=2026-03-13, pubDate=null, pubDateStr=null, doiRegisterDate=null, doiRegisterDateStr=null, onlineIssueDate=1773367977496, onlineIssueDateStr=2026-03-13, onlineJustAcceptDate=null, onlineJustAcceptDateStr=null, onlineFirstDate=null, onlineFirstDateStr=null, sourceXml=null, magXml=null, createTime=1773367977496, creator=13701087609, updateTime=1773367977496, updator=13701087609, issue=Issue{id=1239158367571210992, tenantId=1146029695717560320, journalId=1238841944844054536, year='2025', volume='46', issue='6', pageStart='1', pageEnd='170', issueExtLink='null', onlineDate='null', pubDate='null', beforeIssueId=null, nextIssueId=null, price=null, status=1, issueComplete=1, articleOrder=1, issueType=1, specialIssue=0, createTime=1773367977278, creator=13701087609, updateTime=1773368232261, updator=13701087609, preIssue=null, nextIssue=null, ext={EN=IssueExt(id=1239159437122917323, tenantId=1146029695717560320, journalId=1238841944844054536, issueId=1239158367571210992, language=EN, specialIssueTitle=, coverIllustrator=null, specialIssueEditor=, specialIssueAbout=), CN=IssueExt(id=1239159437122917324, tenantId=1146029695717560320, journalId=1238841944844054536, issueId=1239158367571210992, language=CN, specialIssueTitle=, coverIllustrator=null, specialIssueEditor=, specialIssueAbout=)}, issueFiles=null}, startPage=29, endPage=38, ext={EN=ArticleExt(id=1239158368749801488, articleId=1239158368489754638, tenantId=1146029695717560320, journalId=1238841944844054536, language=EN, title=Multi-Modal Feature Fusion Based Open-Set Modulation Recognition Method for HF Signals, columnId=1239133503154081962, journalTitle=Journal of Telemetry, Tracking and Command, columnName=TT & C Communication and Navigation, runingTitle=null, highlight=null, articleAbstract=

Existing deep learning-based modulation recognition methods are difficult to adapt to HF channels with significant time-varying characteristics, limiting their application in modulation recognition for High Frequency (HF) signals. In addition, in non-cooperative HF communication scenarios, the training dataset is often difficult to cover all possible modulation types, so that open-set modulation recognition also has important practical significance. This paper proposes a multi-modal feature fusion-based open-set modulation recognition method by combining communication domain knowledge and open-set recognition techniques,which effectively reduces the impact of time-varying HF channels and unknown modulation types on recognition performance. The proposed method first utilizes communication domain knowledge to obtain multimodal features that are robust to channel variations,and then extracts discriminative deep feature representations through multimodal feature fusion and deep feature learning to effectively identify known and unknown modulation types. In addition, the method also generates dummy samples through manifold mixing strategy to assist network training, which can enhance the network's ability to identify unknown types. Experimental results indicate that the proposed method outperforms existing open-set modulation recognition methods. When the channel conditions of training and testing signals are the same, the proposed method improves by over 3% in open-set recognition performance. When the channel condition of testing signals is drastically changed, that is, the channel conditions of the training and testing signals are different,the proposed method improves by over 8% compared to existing method, which exhibits strong robustness to channel variations.

, correspAuthors=null, authorNote=null, correspAuthorsNote=null, copyrightStatement=null, copyrightOwner=null, extLink=null, articleAbsUrl=null, sourceXml=null, magXml=null, pdfUrl=null, pdf=null, pdfFileSize=null, pdfExtLink=null, richHtmlUrl=null, mobilePdfUrl=null, reviewReport=null, pdfFirstPage=null, abstractGraph=null, abstractGraphContent=null, abstractVideo=null, citation=null, cebUrl=null, magXmlContent=null, mapNumber=null, authorCompany=null, fund=null, authors=null, authorsList=Huang LIN, Xuchu DAI), CN=ArticleExt(id=1239158381626315150, articleId=1239158368489754638, tenantId=1146029695717560320, journalId=1238841944844054536, language=CN, title=基于多模态特征融合的短波信号开集调制识别方法, columnId=1239133503338631357, journalTitle=遥测遥控, columnName=测控通信与导航, runingTitle=null, highlight=null, articleAbstract=

现有基于深度学习的调制识别方法难以适应具有显著时变特性的短波信道,限制了其在短波信号调制识别中的应用。此外,在非合作短波通信场景中,训练数据集通常难以覆盖所有可能出现的调制类型,因而开集调制识别具有重要应用价值。本文结合通信领域知识和开集识别技术,提出了一种基于多模态特征融合的短波信号开集调制识别方法,有效缓解了短波信道时变性和未知调制类型对识别性能的影响。所提方法首先利用通信领域知识获得对信道变化具有鲁棒性的多模态特征,随后通过多模态特征融合和深度特征学习来提取具有判别性的深度特征表示,从而有效地识别已知和未知调制类型。此外,还通过流形混合策略生成混淆数据作为虚拟样本辅助网络训练,以提高网络对未知类型的识别能力。实验结果表明:所提方法优于现有的开集调制识别方法。当训练与测试信号的信道条件相同时,其开集识别性能有3%以上的性能提升;当测试信号的信道条件发生剧烈变化,即训练与测试信号的信道条件不同时,其性能较现有方法提高了8%以上,对信道变化表现出较强的鲁棒性。

, correspAuthors=null, authorNote=null, correspAuthorsNote=null, copyrightStatement=null, copyrightOwner=null, extLink=null, articleAbsUrl=null, sourceXml=XUmXWZQZlj3biQtWwLHz6A==, magXml=ebImaLprBKVk3j42m+Ogcw==, pdfUrl=null, pdf=QwJ/602u4Nmo2SkyAcui3A==, pdfFileSize=1805176, pdfExtLink=null, richHtmlUrl=null, mobilePdfUrl=null, reviewReport=null, pdfFirstPage=null, abstractGraph=sX//cJs2hDKB1jIfwvPOGg==, abstractGraphContent=null, abstractVideo=null, citation=null, cebUrl=null, magXmlContent=EWTqzTB6TuEIjnD5btsi/w==, mapNumber=null, authorCompany=null, fund=null, authors=

林皇 1993年生,博士研究生。

戴旭初 1963年生,教授,博士生导师。

, authorsList=林皇, 戴旭初)}, authors=[Author(id=1239158381915722146, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, orderNo=0, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1239158381995413930, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, authorId=1239158381915722146, language=EN, stringName=Huang LIN, firstName=Huang, middleName=null, lastName=LIN, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=University of Science and Technology of China, Hefei 230026, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1239158382079300014, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, authorId=1239158381915722146, language=CN, stringName=林皇, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=中国科学技术大学 合肥 230026, bio={"content":"

林皇 1993年生,博士研究生。

"}, bioImg=null, bioContent=

林皇 1993年生,博士研究生。

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1239158381836030362, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, xref=null, ext=[AuthorCompanyExt(id=1239158381844418971, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, companyId=1239158381836030362, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=University of Science and Technology of China, Hefei 230026, China), AuthorCompanyExt(id=1239158381848613276, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, companyId=1239158381836030362, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=中国科学技术大学 合肥 230026)])]), Author(id=1239158382163186105, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, orderNo=1, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1239158382251266497, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, authorId=1239158382163186105, language=EN, stringName=Xuchu DAI, firstName=Xuchu, middleName=null, lastName=DAI, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=University of Science and Technology of China, Hefei 230026, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1239158382360318408, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, authorId=1239158382163186105, language=CN, stringName=戴旭初, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=中国科学技术大学 合肥 230026, bio={"content":"

戴旭初 1963年生,教授,博士生导师。

"}, bioImg=null, bioContent=

戴旭初 1963年生,教授,博士生导师。

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1239158381836030362, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, xref=null, ext=[AuthorCompanyExt(id=1239158381844418971, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, companyId=1239158381836030362, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=University of Science and Technology of China, Hefei 230026, China), AuthorCompanyExt(id=1239158381848613276, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, companyId=1239158381836030362, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=中国科学技术大学 合肥 230026)])])], keywords=[Keyword(id=1239158382502924752, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, language=EN, orderNo=1, keyword=Modulation recognition), Keyword(id=1239158382599393750, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, language=EN, orderNo=2, keyword=HF communication), Keyword(id=1239158382674891226, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, language=EN, orderNo=3, keyword=Multi-modal features), Keyword(id=1239158382767165917, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, language=EN, orderNo=4, keyword=Open-set recognition), Keyword(id=1239158382842663395, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, language=EN, orderNo=5, keyword=Deep feature learning), Keyword(id=1239158382981075432, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, language=CN, orderNo=1, keyword=调制识别), Keyword(id=1239158383203373550, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, language=CN, orderNo=2, keyword=短波通信), Keyword(id=1239158383316619764, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, language=CN, orderNo=3, keyword=多模态特征), Keyword(id=1239158383413088760, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, language=CN, orderNo=4, keyword=开集识别), Keyword(id=1239158383488586237, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, language=CN, orderNo=5, keyword=深度特征学习)], refs=[Reference(id=1239158387179573880, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, doi=null, pmid=null, pmcid=null, year=2018, volume=15, issue=9, pageStart=1, pageEnd=9, url=null, language=null, rfNumber=[1], rfOrder=0, authorNames=WANG J, DING G, WANG H, journalName=China Communications, refType=null, unstructuredReference=WANG J, DING G, WANG H. HF communications: Past,present, and future[J]. China Communications, 2018, 15(9): 1-9., articleTitle=HF communications: Past,present, and future, refAbstract=null), Reference(id=1239158387263459966, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, doi=null, pmid=null, pmcid=null, year=2005, volume=null, issue=null, pageStart=501, pageEnd=512, url=null, language=null, rfNumber=[2], rfOrder=1, authorNames=GIESBRECHT J E, CLARKE R, ABBOTT D, journalName=null, refType=null, unstructuredReference=GIESBRECHT J E, CLARKE R, ABBOTT D. Modulation recognition for HF signals[C]// Smart Structures, Devices, and Systems II. Bellingham: SPIE, 2005: 501-512., articleTitle=Modulation recognition for HF signals, refAbstract=null), Reference(id=1239158387347346049, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, doi=null, pmid=null, pmcid=null, year=2021, volume=33, issue=12, pageStart=7020, pageEnd=7038, url=null, language=null, rfNumber=[3], rfOrder=2, authorNames=PENG S, SUN S, YAO Y D, journalName=IEEE Transactions on Neural Networks and Learning Systems, refType=null, unstructuredReference=PENG S, SUN S, YAO Y D. A survey of modulation classification using deep learning: Signal representation and data preprocessing[J]. IEEE Transactions on Neural Networks and Learning Systems, 2021, 33(12): 7020-7038., articleTitle=A survey of modulation classification using deep learning: Signal representation and data preprocessing, refAbstract=null), Reference(id=1239158387448009348, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, doi=null, pmid=null, pmcid=null, year=2022, volume=129, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=[4], rfOrder=3, authorNames=ZHANG F, LUO C, XU J, journalName=Digit Signal Processing, refType=null, unstructuredReference=ZHANG F, LUO C, XU J, et al. Deep learning based automatic modulation recognition: Models, datasets, and challenges[J]. Digit Signal Processing, 2022, 129: 103650., articleTitle=Deep learning based automatic modulation recognition: Models, datasets, and challenges, refAbstract=null), Reference(id=1239158387531895434, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, doi=null, pmid=null, pmcid=null, year=2007, volume=1, issue=2, pageStart=137, pageEnd=156, url=null, language=null, rfNumber=[5], rfOrder=4, authorNames=DOBRE O A, ABDI A, BAR-NESS Y, journalName=IET Communications, refType=null, unstructuredReference=DOBRE O A, ABDI A, BAR-NESS Y, et al. Survey of automatic modulation classification techniques: Classical approaches and new trends[J]. IET Communications,2007, 1(2): 137-156., articleTitle=Survey of automatic modulation classification techniques: Classical approaches and new trends, refAbstract=null), Reference(id=1239158387636753039, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, doi=null, pmid=null, pmcid=null, year=1995, volume=47, issue=1, pageStart=55, pageEnd=69, url=null, language=null, rfNumber=[6], rfOrder=5, authorNames=AZZOUZ E E, NANDI A K, journalName=Signal Processing, refType=null, unstructuredReference=AZZOUZ E E, NANDI A K. Automatic identification of digital modulation types[J]. Signal Processing, 1995, 47(1): 55-69., articleTitle=Automatic identification of digital modulation types, refAbstract=null), Reference(id=1239158387712250516, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, doi=null, pmid=null, pmcid=null, year=2016, volume=16, issue=1, pageStart=400, pageEnd=411, url=null, language=null, rfNumber=[7], rfOrder=6, authorNames=HAN L, GAO F, LI Z, journalName=IEEE Transactions on Wireless Communications, refType=null, unstructuredReference=HAN L, GAO F, LI Z, et al. Low complexity automatic modulation classification based on order-statistics[J].IEEE Transactions on Wireless Communications, 2016,16(1): 400-411., articleTitle=Low complexity automatic modulation classification based on order-statistics, refAbstract=null), Reference(id=1239158387812913819, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, doi=null, pmid=null, pmcid=null, year=2000, volume=80, issue=2, pageStart=251, pageEnd=277, url=null, language=null, rfNumber=[8], rfOrder=7, authorNames=MOBASSERI B G, journalName=Signal Processing, refType=null, unstructuredReference=MOBASSERI B G. Digital modulation classification using constellation shape[J]. Signal Processing, 2000, 80(2): 251-277., articleTitle=Digital modulation classification using constellation shape, refAbstract=null), Reference(id=1239158387884216991, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, doi=null, pmid=null, pmcid=null, year=2024, volume=19, issue=null, pageStart=3690, pageEnd=3702, url=null, language=null, rfNumber=[9], rfOrder=8, authorNames=CHEN Z, WANG Z, XU D, journalName=IEEE Transactions on Information Forensics and Security, refType=null, unstructuredReference=CHEN Z, WANG Z, XU D, et al. Learn to defend: Adversarial multi-distillation for automatic modulation recognition models[J]. IEEE Transactions on Information Forensics and Security, 2024, 19: 3690-3702., articleTitle=Learn to defend: Adversarial multi-distillation for automatic modulation recognition models, refAbstract=null), Reference(id=1239158387980685987, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, doi=null, pmid=null, pmcid=null, year=2024, volume=23, issue=7, pageStart=8013, pageEnd=8028, url=null, language=null, rfNumber=[10], rfOrder=9, authorNames=HOU D, LI L, LIN W, journalName=IEEE Transactions on Wireless Communications, refType=null, unstructuredReference=HOU D, LI L, LIN W, et al. CIST: A convolutional transformer framework for automatic modulation recognition by knowledge distillation[J]. IEEE Transactions on Wireless Communications, 2024, 23(7): 8013-8028., articleTitle=CIST: A convolutional transformer framework for automatic modulation recognition by knowledge distillation, refAbstract=null), Reference(id=1239158388047794854, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, doi=null, pmid=null, pmcid=null, year=2023, volume=9, issue=3, pageStart=549, pageEnd=563, url=null, language=null, rfNumber=[11], rfOrder=10, authorNames=ZHANG J, WANG T, FENG Z, journalName=IEEE Transactions on Cognitive Communications and Networking, refType=null, unstructuredReference=ZHANG J, WANG T, FENG Z, et al. Toward the automatic modulation classification with adaptive wavelet network[J]. IEEE Transactions on Cognitive Communications and Networking, 2023, 9(3): 549-563., articleTitle=Toward the automatic modulation classification with adaptive wavelet network, refAbstract=null), Reference(id=1239158388152652461, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, doi=null, pmid=null, pmcid=null, year=2022, volume=26, issue=7, pageStart=1588, pageEnd=1592, url=null, language=null, rfNumber=[12], rfOrder=11, authorNames=ZHANG X, LI T, GONG P, journalName=IEEE Communications Letters, refType=null, unstructuredReference=ZHANG X, LI T, GONG P, et al. Open set recognition of communication signal modulation based on deep learning[J]. IEEE Communications Letters, 2022, 26(7):1588-1592., articleTitle=Open set recognition of communication signal modulation based on deep learning, refAbstract=null), Reference(id=1239158388236538542, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, doi=null, pmid=null, pmcid=null, year=2023, volume=27, issue=3, pageStart=851, pageEnd=855, url=null, language=null, rfNumber=[13], rfOrder=12, authorNames=CHEN Y, XU X, QIN X, journalName=IEEE Communications Letters, refType=null, unstructuredReference=CHEN Y, XU X, QIN X. An open-set modulation recognition scheme with deep representation learning[J].IEEE Communications Letters, 2023, 27(3): 851-855., articleTitle=An open-set modulation recognition scheme with deep representation learning, refAbstract=null), Reference(id=1239158388307841715, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, doi=null, pmid=null, pmcid=null, year=2024, volume=28, issue=3, pageStart=627, pageEnd=631, url=null, language=null, rfNumber=[14], rfOrder=13, authorNames=CHEN Y, ZHU L, JIN C, journalName=IEEE Communications Letters, refType=null, unstructuredReference=CHEN Y, ZHU L, JIN C, et al. Boosting open-set RF signal recognition under low snr condition in feature space[J]. IEEE Communications Letters, 2024, 28(3):627-631., articleTitle=Boosting open-set RF signal recognition under low snr condition in feature space, refAbstract=null), Reference(id=1239158388391727798, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, doi=null, pmid=null, pmcid=null, year=2023, volume=45, issue=11, pageStart=13730, pageEnd=13748, url=null, language=null, rfNumber=[15], rfOrder=14, authorNames=LI T, WEN Z, LONG Y, journalName=IEEE Transactions on Pattern Analysis and Machine Intelligence, refType=null, unstructuredReference=LI T, WEN Z, LONG Y, et al. The importance of expert knowledge for automatic modulation open set recognition[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2023, 45(11): 13730-13748., articleTitle=The importance of expert knowledge for automatic modulation open set recognition, refAbstract=null), Reference(id=1239158388488196794, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, doi=null, pmid=null, pmcid=null, year=1970, volume=18, issue=6, pageStart=792, pageEnd=803, url=null, language=null, rfNumber=[16], rfOrder=15, authorNames=WATTERSON C C, JUROSHEK J R, BENSEMA W D, journalName=IEEE Transactions on Communication Technology, refType=null, unstructuredReference=WATTERSON C C, JUROSHEK J R, BENSEMA W D. Experimental confirmation of an HF channel model[J].IEEE Transactions on Communication Technology,1970, 18(6): 792-803., articleTitle=Experimental confirmation of an HF channel model, refAbstract=null), Reference(id=1239158389952008895, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, doi=null, pmid=null, pmcid=null, year=2020, volume=43, issue=10, pageStart=3614, pageEnd=3631, url=null, language=null, rfNumber=[17], rfOrder=16, authorNames=GENG C, HUANG S, CHEN S, journalName=IEEE Transactions on Pattern Analysis and Machine Intelligence, refType=null, unstructuredReference=GENG C, HUANG S, CHEN S. Recent advances in open set recognition: A survey[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2020, 43(10): 3614-3631., articleTitle=Recent advances in open set recognition: A survey, refAbstract=null), Reference(id=1239158390035894979, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, doi=null, pmid=null, pmcid=null, year=2009, volume=31, issue=5, pageStart=1067, pageEnd=1071, url=null, language=null, rfNumber=[18], rfOrder=17, authorNames=杨琳, 许小东, 路友荣, journalName=电子与信息学报, refType=null, unstructuredReference=杨琳, 许小东, 路友荣, 等. 常见数字通信信号的谱线特征分析[J]. 电子与信息学报, 2009, 31(5): 1067-1071., articleTitle=常见数字通信信号的谱线特征分析, refAbstract=null), Reference(id=1239158390140752584, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, doi=null, pmid=null, pmcid=null, year=2009, volume=31, issue=5, pageStart=1067, pageEnd=1071, url=null, language=null, rfNumber=[18], rfOrder=18, authorNames=YANG Lin, XU Xiaodong, LU Yourong, journalName=Journal of Electronics & Information Technology, refType=null, unstructuredReference=YANG Lin, XU Xiaodong, LU Yourong, et al. Spectrum line feature analysis for digital communication signals[J]. Journal of Electronics & Information Technology, 2009, 31(5): 1067-1071., articleTitle=Spectrum line feature analysis for digital communication signals, refAbstract=null), Reference(id=1239158390266581707, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, doi=null, pmid=null, pmcid=null, year=2016, volume=null, issue=null, pageStart=770, pageEnd=778, url=null, language=null, rfNumber=[19], rfOrder=19, authorNames=HE K, ZHANG X, REN S, journalName=null, refType=null, unstructuredReference=HE K, ZHANG X, REN S, et al. Deep residual learning for image recognition[C]// Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.New York: IEEE, 2016: 770-778., articleTitle=Deep residual learning for image recognition, refAbstract=null), Reference(id=1239158390358856401, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, doi=null, pmid=null, pmcid=null, year=2020, volume=null, issue=null, pageStart=6398, pageEnd=6407, url=null, language=null, rfNumber=[20], rfOrder=20, authorNames=SUN Y, CHENG C, ZHANG Y, journalName=null, refType=null, unstructuredReference=SUN Y, CHENG C, ZHANG Y, et al. Circle loss: A unified perspective of pair similarity optimization[C]// Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. New York: IEEE, 2020:6398-6407., articleTitle=Circle loss: A unified perspective of pair similarity optimization, refAbstract=null), Reference(id=1239158390434353875, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, doi=null, pmid=null, pmcid=null, year=2021, volume=null, issue=null, pageStart=4401, pageEnd=4410, url=null, language=null, rfNumber=[21], rfOrder=21, authorNames=ZHOU D W, YE H J, ZHAN D C, journalName=null, refType=null, unstructuredReference=ZHOU D W, YE H J, ZHAN D C. Learning placeholders for open-set recognition[C]// Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. New York: IEEE, 2021: 4401-4410., articleTitle=Learning placeholders for open-set recognition, refAbstract=null), Reference(id=1239158390522434262, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, doi=null, pmid=null, pmcid=null, year=2019, volume=null, issue=null, pageStart=6438, pageEnd=6447, url=null, language=null, rfNumber=[22], rfOrder=22, authorNames=VERMA V, LAMB A, BECKHAM C, journalName=null, refType=null, unstructuredReference=VERMA V, LAMB A, BECKHAM C, et al. Manifold mixup: Better representations by interpolating hidden states[C]// International Conference on Machine Learning. PMLR. New York: Curran Associates, Inc. and International Machine Learning Society, 2019: 6438-6447., articleTitle=Manifold mixup: Better representations by interpolating hidden states, refAbstract=null), Reference(id=1239158390614708957, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, doi=null, pmid=null, pmcid=null, year=2000, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=[23], rfOrder=23, authorNames=International Telecommunication Union, journalName=null, refType=null, unstructuredReference=International Telecommunication Union. Recommendation ITU-R. F. 1487: Testing of HF modems with bandwidths of up to about 12 kHz using ionospheric channel simulators[S]. Geneva: ITU, 2000., articleTitle=null, refAbstract=null), Reference(id=1239158390702789343, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, doi=null, pmid=null, pmcid=null, year=2022, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=[24], rfOrder=24, authorNames=田伟, journalName=航空短波数据通信基带信号处理, refType=null, unstructuredReference=田伟. 航空短波数据通信基带信号处理[M]. 北京: 清华大学出版社, 2022., articleTitle=null, refAbstract=null), Reference(id=1239158390774092517, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, doi=null, pmid=null, pmcid=null, year=2016, volume=null, issue=null, pageStart=1563, pageEnd=1572, url=null, language=null, rfNumber=[25], rfOrder=25, authorNames=BENDALE A, BOULT T E, journalName=null, refType=null, unstructuredReference=BENDALE A, BOULT T E. Towards open set deep networks[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. New York: IEEE,2016: 1563-1572., articleTitle=Towards open set deep networks, refAbstract=null)], funds=[Fund(id=1239158387066327668, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, awardId=61971391, language=CN, fundingSource=国家自然科学基金(61971391), fundOrder=null, country=null)], companyList=[AuthorCompany(id=1239158381836030362, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, xref=null, ext=[AuthorCompanyExt(id=1239158381844418971, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, companyId=1239158381836030362, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=University of Science and Technology of China, Hefei 230026, China), AuthorCompanyExt(id=1239158381848613276, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, companyId=1239158381836030362, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=中国科学技术大学 合肥 230026)])], figs=[ArticleFig(id=1239158383626998280, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, language=EN, label=Fig. 1, caption=Structure of proposed method, figureFileSmall=ZYEE5sFi2a83By8C2GjHBw==, figureFileBig=sX//cJs2hDKB1jIfwvPOGg==, tableContent=null), ArticleFig(id=1239158383777993230, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, language=CN, label=图1, caption=所提方法的基本结构, figureFileSmall=ZYEE5sFi2a83By8C2GjHBw==, figureFileBig=sX//cJs2hDKB1jIfwvPOGg==, tableContent=null), ArticleFig(id=1239158384038040090, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, language=EN, label=Fig. 2, caption=Structure of residual convolutional layer, figureFileSmall=4VJj6WpClL5QckL2Im/gkQ==, figureFileBig=M+dNeaNQ7L4VNyRh5nbKcg==, tableContent=null), ArticleFig(id=1239158385535406625, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, language=CN, label=图2, caption=残差卷积层的结构, figureFileSmall=4VJj6WpClL5QckL2Im/gkQ==, figureFileBig=M+dNeaNQ7L4VNyRh5nbKcg==, tableContent=null), ArticleFig(id=1239158385640264230, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, language=EN, label=Fig. 3, caption=Diagram of manifold mixing strategy, figureFileSmall=6iXbuy7EbMTvZpR1+itdjw==, figureFileBig=/fFvnC6I5wL1fYf4WrbbCA==, tableContent=null), ArticleFig(id=1239158385715761707, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, language=CN, label=图3, caption=流形混合策略的示意图, figureFileSmall=6iXbuy7EbMTvZpR1+itdjw==, figureFileBig=/fFvnC6I5wL1fYf4WrbbCA==, tableContent=null), ArticleFig(id=1239158385791259185, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, language=EN, label=Fig. 4, caption=Comparison of recognition performance under different SNRs, figureFileSmall=4I03xyIuXf9QS1nDc1L0wQ==, figureFileBig=DxjY/7kPyryrKOf4OOgKSw==, tableContent=null), ArticleFig(id=1239158385870950968, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, language=CN, label=图4, caption=不同信噪比下识别性能对比, figureFileSmall=4I03xyIuXf9QS1nDc1L0wQ==, figureFileBig=DxjY/7kPyryrKOf4OOgKSw==, tableContent=null), ArticleFig(id=1239158385933865533, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, language=EN, label=Table 1, caption=

Detailed parameters of HFMS dataset

, figureFileSmall=null, figureFileBig=null, tableContent=
数据集HFMS数据集
信道模型Watterson信道模型
纬度条件低纬度、高纬度
电离层状态安静条件、扰动条件
信噪比范围0 dB~20 dB,间隔2 dB
信号格式IQ复信号,2×9600
信号样本总数1 320 000个信号
调制类型数字调制:BPSK,QPSK,8PSK,8QAM,16QAM,OQPSK,GFSK,CPFSK,OOK,4PAM,8PAM,2FSK,4FSK,8FSK,16FSK
模拟调制:PM,WB-FM,AM-SSB,AM-DSB-WC,AM-DSB-SC
), ArticleFig(id=1239158386005168702, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, language=CN, label=表1, caption=

HFMS数据集的详细参数

, figureFileSmall=null, figureFileBig=null, tableContent=
数据集HFMS数据集
信道模型Watterson信道模型
纬度条件低纬度、高纬度
电离层状态安静条件、扰动条件
信噪比范围0 dB~20 dB,间隔2 dB
信号格式IQ复信号,2×9600
信号样本总数1 320 000个信号
调制类型数字调制:BPSK,QPSK,8PSK,8QAM,16QAM,OQPSK,GFSK,CPFSK,OOK,4PAM,8PAM,2FSK,4FSK,8FSK,16FSK
模拟调制:PM,WB-FM,AM-SSB,AM-DSB-WC,AM-DSB-SC
), ArticleFig(id=1239158386089054786, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, language=EN, label=Table 2, caption=

Specific parameter settings of proposed method

, figureFileSmall=null, figureFileBig=null, tableContent=
参数
边际松弛裕度(m0.25
特征向量维度(d512
尺度因子(γc128
类内样本数目(Np32
类间样本数目(Nn64
虚拟分类器权重(β0.1
分类损失权重(λc1.0
原型学习损失权重(λp0.5
虚拟样本损失权重(λu0.5
), ArticleFig(id=1239158386172940872, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, language=CN, label=表2, caption=

所提方法的具体参数设置

, figureFileSmall=null, figureFileBig=null, tableContent=
参数
边际松弛裕度(m0.25
特征向量维度(d512
尺度因子(γc128
类内样本数目(Np32
类间样本数目(Nn64
虚拟分类器权重(β0.1
分类损失权重(λc1.0
原型学习损失权重(λp0.5
虚拟样本损失权重(λu0.5
), ArticleFig(id=1239158386252632650, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, language=EN, label=Table 3, caption=

Experimental results of different methods

, figureFileSmall=null, figureFileBig=null, tableContent=
方法CCAAUROCOSCR
DEVT91.5%86.0%83.3%
FSSL92.0%87.4%84.6%
MMPRF92.2%90.1%87.1%
本文方法94.3%92.8%90.5%
), ArticleFig(id=1239158386323935823, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, language=CN, label=表3, caption=

不同方法的实验结果

, figureFileSmall=null, figureFileBig=null, tableContent=
方法CCAAUROCOSCR
DEVT91.5%86.0%83.3%
FSSL92.0%87.4%84.6%
MMPRF92.2%90.1%87.1%
本文方法94.3%92.8%90.5%
), ArticleFig(id=1239158386420404820, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, language=EN, label=Table 4, caption=

Experimental results under different channel conditions

, figureFileSmall=null, figureFileBig=null, tableContent=
方法低纬度
扰动条件
高纬度
安静条件
高纬度
扰动条件
DEVT79.4%81.6%76.0%
FSSL80.6%82.8%77.4%
MMPRF83.6%85.4%80.7%
本文方法87.7%89.1%85.2%
), ArticleFig(id=1239158386533651032, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, language=CN, label=表4, caption=

不同信道条件下的实验结果

, figureFileSmall=null, figureFileBig=null, tableContent=
方法低纬度
扰动条件
高纬度
安静条件
高纬度
扰动条件
DEVT79.4%81.6%76.0%
FSSL80.6%82.8%77.4%
MMPRF83.6%85.4%80.7%
本文方法87.7%89.1%85.2%
), ArticleFig(id=1239158386621731422, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, language=EN, label=Table 5, caption=

Effect of channel condition changes on recognition performance

, figureFileSmall=null, figureFileBig=null, tableContent=
方法LQ→HDHD→LQ
AUROCOSCRAUROCOSCR
DEVT70.2%67.1%78.1%75.8%
FSSL71.5%68.7%78.7%76.7%
MMPRF74.8%72.4%82.8%79.8%
本文方法83.7%81.1%89.6%87.1%
), ArticleFig(id=1239158386714006115, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, language=CN, label=表5, caption=

信道条件变化对识别性能的影响

, figureFileSmall=null, figureFileBig=null, tableContent=
方法LQ→HDHD→LQ
AUROCOSCRAUROCOSCR
DEVT70.2%67.1%78.1%75.8%
FSSL71.5%68.7%78.7%76.7%
MMPRF74.8%72.4%82.8%79.8%
本文方法83.7%81.1%89.6%87.1%
), ArticleFig(id=1239158386831446633, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, language=EN, label=Table 6, caption=

Comparison of ablation experiment results

, figureFileSmall=null, figureFileBig=null, tableContent=
方法CCAAUROCOSCR
IQ-MRN92.7%90.7%88.1%
SDC-MRN93.8%91.7%89.3%
NGU-MRN93.6%91.5%88.7%
本文方法94.3%92.8%90.5%
), ArticleFig(id=1239158386957275760, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1239158368489754638, language=CN, label=表6, caption=

消融实验结果对比

, figureFileSmall=null, figureFileBig=null, tableContent=
方法CCAAUROCOSCR
IQ-MRN92.7%90.7%88.1%
SDC-MRN93.8%91.7%89.3%
NGU-MRN93.6%91.5%88.7%
本文方法94.3%92.8%90.5%
)], attaches=null, journal=Journal(id=1238822691516493831, delFlag=0, nameCn=遥测遥控, nameEn=Journal of Telemetry, Tracking and Command, nameHistory1=null, nameHistory2=null, issn=2095-1000, eissn=null, cn=11-1780/TP, coden=null, periodic=1, language=CN, oaType=null, ccby=null, superviseOffice=null, ownerOffice=null, pubOffice=null, editorOffice=null, officeType=null, aims=null, clcCode=null, officeProv=null, officeCity=null, officeAddr=null, officeZip=null, officeEmail=null, officePhone=null, editDirector=null, officeDirector=null, officeDirectorPhone=null, officeStaffNum=null, officeEmpNum=null, coverPicUrl=uUqWBk5pV4VWPULNoAHkuQ==, journalPrice=null, startedYear=null, abbrevIsoEn=Journal of Telemetry, Tracking and Command, journalRemark=null, publicationField=null, createdTime=1773287945871, updatedTime=1773299995823, createdBy=18614031015, updatedBy=13701087609, firstLetterCn=J, firstLetterEn=J, subjectCode=Engineering, subjectName=null, subjectCodeEn=Engineering, subjectNameEn=null, picCn=uUqWBk5pV4VWPULNoAHkuQ==, picEn=RDDslHHvDBDNnJMjk6BxdA==, jcr=null, cjcr=null, exts=[JournalExt(id=1238873232732909900, language=CN, name=遥测遥控, nameHistory1=null, nameHistory2=null, managedBy=, sponsoredBy=, publishedBy=, editorOffice=, officeProv=null, officeCity=null, officeAddr=, officeZip=, editDirector=, officeDirector=null, officePhone=null, coverPicUrl=null, journalRemark=, submitArticleUrl=null, websiteUrl=, createdTime=1773299995835, updatedTime=1773299995835, createdBy=13701087609, updatedBy=13701087609, submissionGuidelinesUrl=, submissionAuthorUrl=https://manuscript.spacejournal.cn/ycyk, submissionEditorUrl=https://manuscript.spacejournal.cn/ycyk, submissionReviewUrl=https://manuscript.spacejournal.cn/ycyk, submissionCeEditorUrl=, submissionAeEditorUrl=, option={"copyright":""}), JournalExt(id=1238873232779047245, language=EN, name=Journal of Telemetry, Tracking and Command, nameHistory1=null, nameHistory2=null, managedBy=, sponsoredBy=, publishedBy=, editorOffice=, officeProv=null, officeCity=null, officeAddr=, officeZip=, editDirector=, officeDirector=null, officePhone=null, coverPicUrl=null, journalRemark=, submitArticleUrl=null, websiteUrl=, createdTime=1773299995846, updatedTime=1773299995846, createdBy=13701087609, updatedBy=13701087609, submissionGuidelinesUrl=, submissionAuthorUrl=https://manuscript.spacejournal.cn/ycyk, submissionEditorUrl=https://manuscript.spacejournal.cn/ycyk, submissionReviewUrl=https://manuscript.spacejournal.cn/ycyk, submissionCeEditorUrl=, submissionAeEditorUrl=, option={"copyright":""})], databaseList=null, tenantJournalId=1238841944844054536, websiteList=[Website(id=1238843830011744952, webName=null, webTitle=null, webDomain=null, webCopyrigh=null, webIpcNo=null, seoTitle=null, seoKeywords=null, seoDescription=null, tenantJournalId=null, journalId=1238841944844054536, journalNameCn=null, journalNameEn=null, grayFlag=null, tenantId=1146029695717560320, platformId=null, journalGroupId=null, journalGroupNameCn=null, journalGroupNameEn=null, type=1, domain=https://castjournals.cast.org.cn/joweb/ycyk/CN, language=CN, createTime=1773292985680, createBy=18614031015, updateTime=1773293577303, updateBy=18614031015, name=遥测遥控-中文, tplId=1146099689490845704, title=遥测遥控, delFlag=0, indexPage=/home, props=[WebsiteProps(id=1238846390617568150, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1238843830011744952, code=articleTextType, value=kx, createTime=1773293596176, updateTime=1773293596176, creator=18614031015, updator=18614031015), WebsiteProps(id=1238846390596596627, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1238843830011744952, code=banner, value=null, createTime=1773293596171, updateTime=1773293596171, creator=18614031015, updator=18614031015), WebsiteProps(id=1238846390638539673, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1238843830011744952, code=grayFlag, value=0, createTime=1773293596181, updateTime=1773293596181, creator=18614031015, updator=18614031015), WebsiteProps(id=1238846390588208018, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1238843830011744952, code=logo, value=https://castjournals.cast.org.cn/joweb/ycyk/CN/file/pic?fileId=g9S2GGWx7iZjf51o9vpmmA==, createTime=1773293596169, updateTime=1773293596169, creator=18614031015, updator=18614031015), WebsiteProps(id=1238846390651122587, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1238843830011744952, code=minRunFlag, value=0, createTime=1773293596184, updateTime=1773293596184, creator=18614031015, updator=18614031015), WebsiteProps(id=1238846390613373845, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1238843830011744952, code=picServerUrl, value=https://castjournals.cast.org.cn/joweb/ycyk/CN/file/pic, createTime=1773293596175, updateTime=1773293596175, creator=18614031015, updator=18614031015), WebsiteProps(id=1238846390646928282, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1238843830011744952, code=silenceFlag, value=0, createTime=1773293596183, updateTime=1773293596183, creator=18614031015, updator=18614031015), WebsiteProps(id=1238846390604985236, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1238843830011744952, code=staticResourcePath, value=https://castjournals.cast.org.cn/joweb/cast_kjdb_cn_619/, createTime=1773293596173, updateTime=1773293596173, creator=18614031015, updator=18614031015), WebsiteProps(id=1238846390625956759, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1238843830011744952, code=themeColor, value=null, createTime=1773293596178, updateTime=1773293596178, creator=18614031015, updator=18614031015), WebsiteProps(id=1238846390634345368, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1238843830011744952, code=themeStyle, value=null, createTime=1773293596180, updateTime=1773293596180, creator=18614031015, updator=18614031015)]), Website(id=1238843830078853819, webName=null, webTitle=null, webDomain=null, webCopyrigh=null, webIpcNo=null, seoTitle=null, seoKeywords=null, seoDescription=null, tenantJournalId=null, journalId=1238841944844054536, journalNameCn=null, journalNameEn=null, grayFlag=null, tenantId=1146029695717560320, platformId=null, journalGroupId=null, journalGroupNameCn=null, journalGroupNameEn=null, type=1, domain=https://castjournals.cast.org.cn/joweb/ycyk/EN, language=EN, createTime=1773292985696, createBy=18614031015, updateTime=1773293566698, updateBy=18614031015, name=遥测遥控-英文, tplId=1146101810881728533, title=Journal of Telemetry, Tracking and Command, delFlag=0, indexPage=/home, props=[WebsiteProps(id=1238846419235304352, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1238843830078853819, code=articleTextType, value=kx, createTime=1773293602999, updateTime=1773293602999, creator=18614031015, updator=18614031015), WebsiteProps(id=1238846419214332829, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1238843830078853819, code=banner, value=null, createTime=1773293602994, updateTime=1773293602994, creator=18614031015, updator=18614031015), WebsiteProps(id=1238846419256275875, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1238843830078853819, code=grayFlag, value=0, createTime=1773293603004, updateTime=1773293603004, creator=18614031015, updator=18614031015), WebsiteProps(id=1238846419205944220, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1238843830078853819, code=logo, value=https://castjournals.cast.org.cn/joweb/ycyk/EN/file/pic?fileId=g9S2GGWx7iZjf51o9vpmmA==, createTime=1773293602992, updateTime=1773293602992, creator=18614031015, updator=18614031015), WebsiteProps(id=1238846419268858789, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1238843830078853819, code=minRunFlag, value=0, createTime=1773293603007, updateTime=1773293603007, creator=18614031015, updator=18614031015), WebsiteProps(id=1238846419231110047, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1238843830078853819, code=picServerUrl, value=https://castjournals.cast.org.cn/joweb/ycyk/EN/file/pic, createTime=1773293602998, updateTime=1773293602998, creator=18614031015, updator=18614031015), WebsiteProps(id=1238846419264664484, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1238843830078853819, code=silenceFlag, value=0, createTime=1773293603006, updateTime=1773293603006, creator=18614031015, updator=18614031015), WebsiteProps(id=1238846419222721438, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1238843830078853819, code=staticResourcePath, value=https://castjournals.cast.org.cn/joweb/cast_kjdb_en_623/, createTime=1773293602996, updateTime=1773293602996, creator=18614031015, updator=18614031015), WebsiteProps(id=1238846419243692961, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1238843830078853819, code=themeColor, value=null, createTime=1773293603001, updateTime=1773293603001, creator=18614031015, updator=18614031015), WebsiteProps(id=1238846419252081570, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1238843830078853819, code=themeStyle, value=null, createTime=1773293603003, updateTime=1773293603003, creator=18614031015, updator=18614031015)])], journalTitle=遥测遥控, weixinUrl=null, journalUrl=https://ycyk.spacejournal.cn/, iacademicId=null, status=1, seqNo=null, journalTitleEn=Journal of Telemetry, Tracking and Command, journalPhotoCn=uUqWBk5pV4VWPULNoAHkuQ==, journalPhotoEn=RDDslHHvDBDNnJMjk6BxdA==, journalFirstLetter=J, journalRecommend=null, journalNew=null, journalCollection=null, jcrJf=null, cjcrJf=null, jcrJfStr=null, cjcrJfStr=null, submissionFirstDecision=null, sciSubjectClassification=null, casSubjectClassification=null, citeScore=null, totalCitationFrequency=null, icpCode=null, psCode=null, advertisingLicenseCode=null, copyrightInformation=null, country=null, option=, provinceCode=null, provinceName=null, collectFlag=false), detailUrlCn=https://castjournals.cast.org.cn/joweb/ycyk/CN/10.12347/j.ycyk.20250723001, detailUrlEn=https://castjournals.cast.org.cn/joweb/ycyk/EN/10.12347/j.ycyk.20250723001, pdfUrlCn=https://castjournals.cast.org.cn/joweb/ycyk/CN/PDF/10.12347/j.ycyk.20250723001, pdfUrlEn=https://castjournals.cast.org.cn/joweb/ycyk/EN/PDF/10.12347/j.ycyk.20250723001, aliStartDate=null, aliEndDate=null, collectionFlag=false, citedCount=null, citedUrl=null, reference=null)
收藏切换
基于多模态特征融合的短波信号开集调制识别方法
收藏切换
PDF下载
林皇 , 戴旭初
遥测遥控 | 测控通信与导航 2025,46(6): 29-38
收起
收藏切换
遥测遥控 | 测控通信与导航 2025, 46(6): 29-38
基于多模态特征融合的短波信号开集调制识别方法
全屏
林皇, 戴旭初
作者信息
  • 中国科学技术大学 合肥 230026
  • 林皇 1993年生,博士研究生。

    戴旭初 1963年生,教授,博士生导师。

Multi-Modal Feature Fusion Based Open-Set Modulation Recognition Method for HF Signals
Huang LIN, Xuchu DAI
Affiliations
  • University of Science and Technology of China, Hefei 230026, China
doi: 10.12347/j.ycyk.20250723001
文章导航
收藏切换

现有基于深度学习的调制识别方法难以适应具有显著时变特性的短波信道,限制了其在短波信号调制识别中的应用。此外,在非合作短波通信场景中,训练数据集通常难以覆盖所有可能出现的调制类型,因而开集调制识别具有重要应用价值。本文结合通信领域知识和开集识别技术,提出了一种基于多模态特征融合的短波信号开集调制识别方法,有效缓解了短波信道时变性和未知调制类型对识别性能的影响。所提方法首先利用通信领域知识获得对信道变化具有鲁棒性的多模态特征,随后通过多模态特征融合和深度特征学习来提取具有判别性的深度特征表示,从而有效地识别已知和未知调制类型。此外,还通过流形混合策略生成混淆数据作为虚拟样本辅助网络训练,以提高网络对未知类型的识别能力。实验结果表明:所提方法优于现有的开集调制识别方法。当训练与测试信号的信道条件相同时,其开集识别性能有3%以上的性能提升;当测试信号的信道条件发生剧烈变化,即训练与测试信号的信道条件不同时,其性能较现有方法提高了8%以上,对信道变化表现出较强的鲁棒性。

调制识别  /  短波通信  /  多模态特征  /  开集识别  /  深度特征学习

Existing deep learning-based modulation recognition methods are difficult to adapt to HF channels with significant time-varying characteristics, limiting their application in modulation recognition for High Frequency (HF) signals. In addition, in non-cooperative HF communication scenarios, the training dataset is often difficult to cover all possible modulation types, so that open-set modulation recognition also has important practical significance. This paper proposes a multi-modal feature fusion-based open-set modulation recognition method by combining communication domain knowledge and open-set recognition techniques,which effectively reduces the impact of time-varying HF channels and unknown modulation types on recognition performance. The proposed method first utilizes communication domain knowledge to obtain multimodal features that are robust to channel variations,and then extracts discriminative deep feature representations through multimodal feature fusion and deep feature learning to effectively identify known and unknown modulation types. In addition, the method also generates dummy samples through manifold mixing strategy to assist network training, which can enhance the network's ability to identify unknown types. Experimental results indicate that the proposed method outperforms existing open-set modulation recognition methods. When the channel conditions of training and testing signals are the same, the proposed method improves by over 3% in open-set recognition performance. When the channel condition of testing signals is drastically changed, that is, the channel conditions of the training and testing signals are different,the proposed method improves by over 8% compared to existing method, which exhibits strong robustness to channel variations.

Modulation recognition  /  HF communication  /  Multi-modal features  /  Open-set recognition  /  Deep feature learning
林皇, 戴旭初. 基于多模态特征融合的短波信号开集调制识别方法. 遥测遥控, 2025 , 46 (6) : 29 -38 . DOI: 10.12347/j.ycyk.20250723001
Huang LIN, Xuchu DAI. Multi-Modal Feature Fusion Based Open-Set Modulation Recognition Method for HF Signals[J]. Journal of Telemetry, Tracking and Command, 2025 , 46 (6) : 29 -38 . DOI: 10.12347/j.ycyk.20250723001
短波通信,又称为高频(High Frequency,HF)通信,主要依靠电离层反射进行中远距离通信,其凭借通信距离远、抗毁能力强、机动性高等优势,在军事通信、海上通信、应急通信等领域中发挥着不可替代的作用[1]。自动调制识别(Automatic Modulation Recognition,AMR)作为短波侦察系统的关键技术之一,其旨在非合作通信场景中分析接收信号所使用的调制以及调制阶数,是后续信号解调解码的基础。相比于其他频段,由于短波传输信道的复杂性,短波信号调制识别更具有挑战性[2]
现有的AMR方法主要依靠从接收信号中提取各种特征实现,根据特征提取的方式不同可以分为基于传统特征的方法和基于深度学习的方法[3-5]。基于传统特征的方法依靠领域专家的专业知识设计的各种信号特征以区分不同的调制类型,常用特征包括信号参数特征[6]、信号高阶统计特征[7]以及变换域特征[8]等。基于深度学习的方法通过深度神经网络自动从大量信号样本中学习不同调制类型的具有区分度的深度特征以进行识别。随着深度学习技术的崛起,调制识别的研究逐渐从“知识驱动”转向“数据驱动”,基于深度学习的方法逐渐成为AMR领域的研究热点,研究人员已经提出了许多方法并取得了不错的识别效果[9-11]
现有的AMR方法大多考虑闭集设置的调制识别,即假定在训练阶段已知所有可能的调制类型。然而,在实际的场景中,测试阶段可能会出现训练阶段中未见过的调制类型。因此,近年来,研究人员对开集调制识别(Open-Set Modulation Recognition,OSMR)问题开展了研究。文献[12]使用改进的端到端损失函数训练神经网络,通过学习不同调制类型的相似性,实现了开集调制类型识别。文献[13]通过极值理论对训练信号与聚类中心的距离进行建模,并使用多个二元分类器对已知和未知调制类型进行区分。文献[14]提出了基于特征空间奇异性的调制识别框架,其鼓励为每种调制类型形成紧凑的分类边界,同时将已知类型与初始噪声推开,从而减少已知和未知类型的识别错误。文献[15]提出了一种多模态边缘原型框架用于射频信号的开集识别,其通过一对一和边际限制的方式划分特征空间以解决开集识别问题。以上的OSMR方法主要是通过现有的开集识别方法改进得到的,对通信领域的相关知识少有考虑,这些方法仍存在进一步提升的空间。
为解决短波信号开集调制识别问题,本文将通信领域知识与开集识别技术相结合,提出一种基于多模态特征融合的调制识别方法。利用特征提取模块,本文所提方法首先通过专家知识和信号处理技术提取接收信号不同表征域中的信号特征(包括瞬时特征和谱特征)。瞬时特征能够反映不同调制类型的瞬时幅度、相位和频率等信息,而谱特征展现了不同调制信号的高阶谱特性。随后,通过特征融合模块提取不同模态的深度特征表示并进行融合。最后,通过调制识别模块学习不同调制类型具有鉴别性的深度特征表示,从而减轻由未知信号引起的识别错误。本文通过仿真的短波调制信号数据集进行广泛的实验验证。实验结果表明:相比于现有的OSMR方法,所提方法能够实现最先进的开集调制识别性能。
在短波通信系统中,信息通常会被编码并调制成无线电信号进行传输。通常,通过短波窄带接收机经过下变频处理得到基带接收信号xt)可以表示为:
式中,f0ϕ0分别表示载波频率偏移和相位偏移,st)表示传输的调制信号,表示经过传输信道后得到的传输信号,zt)为加性噪声,htτ)表示短波传输信道的幅度响应,通常采用Watterson信道模型[16]。无论是模拟还是数字调制,调制信号都是通过幅度、相位和频率的变化将需要传递的信息携带在载波上进行传输的:
式中,At)、ϕt)和Ft)分别表示调制幅度、相位和频率。
为了便于计算机处理,接收方通常会对接收信号进行数字化处理。假设接收机的采样率为fs,则离散的接收信号可以表示为,其中n=0,1,...,N-1。为简单起见,使用复数值向量xCN表示接收信号。本文中信噪比(Signal-to-Noise Ratio,SNR)定义为传输信号的平均功率与加性噪声的平均功率之比,其可以表示为:
其中,r[n]和z[n]分别表示rt)和zt)经过采样后的离散样本,N表示离散采样点数。
基于深度学习的调制识别方法通常只能识别在训练阶段出现过的调制类型。然而,在实际短波调制识别场景中,可能出现未知调制类型(未在训练数据集中出现)。因此,短波信号开集调制识别问题不仅需要识别短波频段内属于已知的各个调制类型的信号,还需要判断信号是否来自未知调制类型,其可以被建模为深度学习中OSR(开集识别)[17]问题,定义如下:
给定从短波电磁环境中采集的有标签的训练数据集DTR={(x1y1),...,(xNyN)}和已知调制类型集合Cs={c1,…,cK},其中yiCs表示接收信号xi的标签,N表示训练信号的数目,K表示已知调制类型的数目。然而,在测试数据集中,既包含来自已知调制类型集合Cs的测试信号xci,也包含来自未知调制类型集合O={o1,…,oU}的测试信号xoi。开集调制识别的目标是训练一个调制识别模型h,使得其对于任意测试信号xjDTE满足:
h能够准确分类属于Cs的测试信号xci
h能够将属于O的测试信号xoi识别为“未知”类型。
根据以上的问题定义,模型只需要将xoi标记为“未知”类型,并不要求将其精确分类到具体的未知调制类型。为了简化问题,可以将所有未知调制类型统一标记为“未知”类型。不失一般性的,可以使用Ct={c1,…,cKcK+1}表示测试数据集的调制类型集合,其中cK+1表示未知类型。短波信号开集调制识别问题的难点在于,在训练数据集中仅包含属于已知调制类型的信号,而未知类型在训练过程中不会被模型训练并学习。因此,需要为已知调制类型形成紧凑的特征空间并利用已有数据模仿未知类型出现的情况,从而更好地区分已知和未知调制类型。
与现有的AMR方法不同,本文不直接使用接收信号的同相和正交(In-phase and Quadrature,IQ)数据,而是采用其多模态的特征表示进行识别,以减轻短波传输信道对调制识别的影响。所采用的多模态特征包括:瞬时特征和谱特征。具体的,接收信号x[n]的瞬时特征包括瞬时幅度Ax[n]和瞬时频率Fx[n],其能够直接反映不同调制类型的瞬时幅度、相位和频率等信息:
式中,ϕx[n]=arctan(xQ[n]/xI[n])表示x[n]的瞬时相位,xI[n]和xQ[n]分别表示x[n]的实部与虚部。
谱特征利用调制信号的循环平稳特性通过接收信号的高阶谱特性反映调制信息,可以有效抵抗短波传输信道中多径效应和衰落的影响[18]
式中,X1[k]、X2[k]和X4[k]分别表示x[n]的频谱、平方谱和四次方谱。
根据表征域的不同,可以将以上特征分为两个不同的模态:
式中,M1M2分别表示接收信号的瞬时特征和谱特征。接收信号不同模态的特征表示可以很好地反映不同调制类型的信号特性,为处理复杂短波电磁环境下的调制识别问题提供了坚实的基础。
通过结合通信领域知识和深度学习技术,本文提出了基于多模态特征融合的调制识别方法。所提方法的基本结构如图1所示,其由三个部分组成:特征提取模块E、特征融合模块F以及调制识别模块R
特征提取模块通过数字信号处理技术从接收信号中提取多模态特征,为后续的信号调制识别提供多模态信息。模块采用多种数据增强的方法扩展训练信号的样本空间,以提高模型的泛化能力,包括添加随机频移、高斯噪声、相位旋转等。这些方法不会影响信号的调制类型,可以增加训练样本的多样性,提高调制识别性能。
特征融合模块能够将不同模态的特征通过神经网络进行融合,其主要由前置层和后置层组成。在前置层中,模块首先通过级联的残差卷积层提取瞬时特征和谱特征的深度特征表示f1=F1M1)和f2=F2M2),随后利用拼接层将提取的深度特征表示进行拼接得到fM=Concat(f1f2),其中残差卷积层采用跨层的残差连接机制的卷积神经网络[19],其结构如图2所示。在后置层中,模块通过残差卷积层和全连接层将fM映射到高维特征空间中得到融合特征向量vfRd×1,其中d表示特征向量的维度。而特征融合模块可以分解为前置层和后置层,其可以表示为:
其中,FpreFpost分别表示前置层和后置层。
调制识别模块利用共享-虚拟原型分类器对网络模型进行训练学习,使得其不仅能对已知调制类型进行准确分类,还能有效区分已知和未知调制类型。共享-虚拟分类器由共享分类器RS和虚拟分类器RD组成,其可以通过深度度量学习和原型学习对提取的融合特征向量进行学习。
为了使得具有相同调制类型的信号的特征表示在特征空间中更加紧凑,以区分已知和未知调制类型,调制识别模块通过深度度量学习对提取的深度特征进行学习。给定从训练数据集DTR中随机采样的训练信号(xy),并随机选择Np个具有相同类型的信号记为类内样本和Nn个不同类型的信号记为类间样本。本文使用圆形损失(Circle Loss)[20]对提取的特征向量进行学习,其可以表示为:
式中,γc为尺度因子,分别表示类内和类间相似度。
式中,(•)T表示向量转置操作,‖•‖2表示L2-范数,vf表示x通过网络得到融合特征向量,分别表示通过网络得到的融合特征向量。表示相似度的权重因子。本文遵循圆形损失的参数设置,Δp=1-m,Δn=mOp=1+mOn=-m,其中m∈[0,1)表示边际松弛裕度。通过深度度量学习,网络能够提取接收信号具有鉴别性的深度特征表示,从而更好地区分不同的调制类型。
共享-虚拟分类器旨在模仿未知类型出现的情况,从而将闭集训练转变为开集训练,并通过虚拟分类器分离已知和未知类型。共享分类器RS和虚拟分类器RD的权重矩阵为:
式中,表示第k个已知调制类型的原型向量,表示虚拟类型的原型向量,U为预设的虚拟类型的数目。因此,共享-虚拟分类器的预测概率向量可以表示为:
式中,σ表示soft max函数,表示虚拟分类器与vf的最大相似度得分,pcx)∈[0,1]K+1表示预测概率向量。由文献[21]可知,通过强制虚拟分类器为已知类型输出第二大的概率,可以有效避免共享分类器由于过度自信引起的错误识别。因此,共享-虚拟分类器的分类损失可以定义为:
式中,pcx)\y∈[0,1]K表示去除输入信号x的真实标签y对应预测后的输出概率,表示交叉熵损失函数,β表示虚拟分类器的权重。通过共享-虚拟分类器的分类损失,可以控制虚拟类型的原型在已知和未知类型之间,当出现新的未知类型时,其将在虚拟类型上输出较大的概率。
共享-虚拟分类器的另一个目标是使得属于已知调制类型的特征向量更接近对应原型向量,从而形成紧凑的特征空间以区分不同的类型,则原型学习损失可以定义为:
式中,表示标签y对应的原型向量。原型学习损失通过优化原型向量和特征向量之间的距离,可以使得已知类型靠近对应的原型向量,从而减少不同调制类型之间的识别错误。
虽然共享-虚拟分类器能够在没有先验知识的情况下区分未知类型,但其仍容易受到未知数据的影响。为进一步减少未知数据引起的识别错误,本文采用流形混合策略[22]生成混淆数据作为虚拟样本,以提高模型对未知调制类型的区分能力。基于流形混合策略的虚拟样本生成的具体实现过程如图3所示。
首先,从训练样本中随机选择具有不同调制类型的两个信号xixj,并通过前置层提取相应的深度特征。随后,对两个信号所提取深度特征进行加权混合:
其中,λ∈(0,1)表示混合系数由Beta分布采样得到,表示混合的特征嵌入。最后,将通过Fpost得到混合特征向量。考虑到不同类型的混合特征向量通常处于低置信度的区域,即非已知类型区域。因此,可以将作为虚拟样本的特征向量进行学习,则虚拟样本损失可以定义为:
其中,表示调制识别模块对虚拟样本的预测概率。
综上所述,可以使用以下的损失函数对网络参数进行更新:
式中,λcλpλu分别表示各个损失的权重系数。
本文采用仿真生成的短波调制信号(High Frequency Modulation Signal,HFMS)数据集进行实验,以评估不同方法在短波电磁环境下开集调制识别性能。该数据集使用Watterson信道模型,以模拟短波信号经过电离层反射进行传播的过程,其根据不同纬度和电离层状态可以获得不同短波信道条件。该数据集中考虑了四种典型的短波信道条件,包括低纬度安静条件(Low latitudes Quiet conditions)、低纬度扰动条件(Low latitudes Disturbed conditions)、高纬度安静条件(High latitudes Quiet conditions)以及高纬度扰动条件(High latitudes Disturbed conditions)。这些短波信道条件具有不同的多径时延、衰落特性以及时变特性等,具体信道参数参见文献[23]。该数据集中包含二十种常见的短波调制类型,包含数字调制和模拟调制两类。参考实际的短波通信应用[24],数字调制方式包括不同阶数的频移键控(Frequency Shift Keying,FSK)、相移键控(Phase Shift Keying,PSK)和正交幅度调制(Quadrature Amplitude Modulation,QAM)等,其符号速率范围从100波特到2 400波特;模拟调制方式包括幅度调制(Amplitude Modulation,AM)和相位调制(Phase Modulation,PM)等,调制信号为实际的话音数据。该数据集中每种调制类型的信噪比范围为0 dB到20 dB,其采样率为9 600 Hz,每个信号样本包含9 600个采样点。在训练和测试数据集中,每种调制类型在每个信噪比下分别包含4 000个和2 000个信号样本。HFMS数据集的具体参数如表1所示。
实验所使用的硬件平台为64 GB内存的Intel Core i9-9900K CPU和两个NVIDIA 2080Ti GPU显卡。训练数据集与测试数据集均是使用MATLAB R2021a仿真生成,深度学习模型通过Python 3.9中PyTorch 1.13.1上实现。所提方法使用适应性矩估计(Adaptive moment estimation,Adam)优化器进行训练,初始学习率为0.001,权重衰减为0.000 1。在开集调制识别场景设置中,数据集中所有调制类型被随机分为已知或未知调制类型之一。所提方法的具体参数设置如表2所示。
本文使用开放度O*衡量开集调制识别问题的开放性[25],其定义为:
其中,MTRMTE分别表示训练数据集和测试数据集中调制类型的数目。通常,较高的O*值意味着测试数据集中包含更多的未知调制类型,因而开集调制识别问题更具挑战性。
与闭集调制识别不同,开集调制识别不仅需要准确识别已知调制类型,还需要区分已知和未知类型,因而闭集调制识别的评价指标不再适用。参考现有的OSMR方法[13-15],本文使用闭集分类准确率(Closed-set Classification Accuracy,CCA)、接收机运行特征曲线下面积(Area Under the Receiver Operating Characteristic curve,AUROC)和开集分类准确率(Open-Set Classification Rate,OSCR)三个指标对不同方法的开集调制识别性能进行比较。在这三个指标中,CCA侧重于衡量算法的闭集识别性能,而AUROC和OSCR侧重于衡量算法的开集识别性能,其中AUROC仅考虑已知和未知调制类型区分性,而OSCR进一步评估了算法对已知类别预测的准确性。
为模拟实际短波信号开集调制识别场景,本节使用低纬度安静条件下的HFMS数据集进行实验,所提方法的实验参数如表2所示。实验选择十五种调制类型作为已知调制类型,包括BPSK,QPSK,8PSK,8QAM,16QAM,OQPSK,GFSK,CPFSK,OOK,4PAM,8PAM,2 FSK,4FSK,WB-FM和AM-SSB,其余作为未知调制类型,此时开放度O*为7.4%。本节将所提方法与现有的OSMR方法进行比较,以评估其在开集场景下短波调制识别性能。主要对比方法包括:基于深度特征学习和极值理论的DEVT[13]、基于特征空间奇异性学习的FSSL[14]和基于多模态特征和对抗性边际原型学习的MMPRF[15],对比方法的实验参数设置遵循其默认设置。表3中展示了所提方法与现有方法的开集调制识别性能对比。
实验结果表明:相比于现有方法,所提方法能够取得最佳的开集调制识别性能。所提方法实现了约94.3%的CCA性能,这表明所提方法可以实现良好的闭集调制识别性能,能够有效解决已知调制类型识别问题。此外,相比于DEVT、FSSL和MMPRF,所提方法分别提高了约6.8%、5.4%和2.7%的AUROC性能和7.2%、5.9%和3.4%的OSCR性能。对比方法中,DEVT、FSSL直接使用IQ数据作为输入,MMPRF虽然采用了接收信号的多个特征,但只包含了IQ数据以及瞬时幅度、频率和相位数据,这些信号特征仍然容易受到短波信道的影响。
进一步,图4中展示了不同方法的开集识别性能随信噪比变化的曲线,以分析信噪比对于各个方法识别性能的影响。实验结果表明:随着信噪比的增加,所有方法的AUROC性能和OSCR性能也随之提高,表明较高的信噪比有助于提升调制识别的准确性。值得注意的是,所提方法在所有信噪比范围内都取得了最佳的开集调制识别性能,尤其是所提方法在信噪比为10 dB时实现了约95%的AUROC性能和93%的OSCR性能,并且其在低信噪比时明显优于对比方法。以上实验结果表明了所提方法对于复杂多变的短波电磁环境具有良好的适应能力与实际应用价值。
为分析不同短波信道条件对不同方法识别性能的影响,本节将分别使用低纬度扰动条件、高纬度安静条件和高纬度扰动条件下的HFMS数据集进行实验。表4中给出了在不同信道条件下各个方法的OSCR性能对比。
实验结果表明:与现有方法相比,所提方法在不同信道条件下都能够取得最佳的识别性能,其在低纬度扰动条件、高纬度安静条件和高纬度扰动条件下相比于对比方法分别提高了约4.1%、3.7%和4.5%的OSCR性能。在这些信道条件中,高纬度扰动条件是最为复杂的短波电磁环境,其具有较大的多径时延、多普勒频移以及快衰落,可以看作快时变信道。因此,相较对其他信道条件,所提方法在高纬度扰动条件下性能提升更加明显,并且其实现了85%以上的OSCR性能。以上实验结果表明了所提方法对于不同信道条件的短波电磁环境的鲁棒性,能够在不同短波信道条件下都实现优异的调制识别性能。
此外,由于短波传输信道为时变信道,训练与测试信号可能采集自不同信道条件的短波电磁环境,从而导致二者的数据分布存在差异。为评估信道条件变化对不同方法的识别性能的影响,本节分别使用低纬度安静条件和高纬度扰动条件下的HFMS数据集,交替作为训练集和测试集进行实验。实验结果如表5所示,其中“→”的两边分别表示训练集和测试集对应的信道条件,LQ表示低纬度安静条件,HD表示高纬度扰动条件。实验结果表明:当信道条件发生变化时,所有方法的识别性能都会出现不同程度的下降,但所提方法仍优于对比方法。当使用低纬度安静条件作为训练集,而高纬度扰动条件作为测试集时,所提方法相较对比方法提高了约8.9%的AUROC性能和8.7%的OSCR性能;当使用高纬度扰动条件作为训练集,而低纬度安静条件作为测试集时,所提方法相较对比方法提高了约6.8%的AUROC性能和7.4%的OSCR性能。这些实验结果表明了所提方法能够有效缓解短波信道的时变性对识别性能的影响。
为评估所提方法中不同组件对调制识别性能的影响,本节通过消融实验将所提方法与其变体方法进行比较,以展示每个组件的对识别性能的贡献。每个变体方法通过单独移除所提方法中的一个组件以衡量其在整体方法中的性能贡献,其中包括移除了多模态特征提取的基于IQ数据的调制识别网络(IQ-based Modulation Recognition Network,IQ-MRN)方法、移除了深度度量学习损失的基于虚拟共享分类器的调制识别网络(Shared-Dummy Classifier-based Modulation Recognition Network,SDC-MRN)以及移除了虚拟样本生成的基于无未知类型生成的调制识别网络(No Generation for Unknown type-based Modulation Recognition Network,NGU-MRN)方法。实验在低纬度安静条件下的HFMS数据集上进行,除了移除的组件部分,各个变体方法的其余实验参数设置遵循表2中参数设置。实验中选择十五种调制类型作为已知调制类型,其余作为未知调制类型,此时开放度O*为7.4%。表6中展示了不同方法的对比实验结果。
实验结果表明:所提方法优于其他的对比方法,能够实现最佳的开集调制识别性能。相比于IQ-MRN,所提方法的CCA性能、AUROC性能和OSCR性能分别提高了约1.5%、2.1%和2.4%,表明了多模态特征对于短波信号调制识别的有效性。SDC-MRN的各个性能也出现了不同程度的下降,表明通过深度度量学习可以有效地提高开集识别性能。与所提方法相比,NGU-MRN的识别性能出现了明显下降,其AUROC性能和OSCR性能分别下降了约1.3%和1.8%,表明了使用虚拟样本生成对于识别未知类型的重要性。这些实验结果验证了所提方法中每个组件对于短波信号开集调制识别的重要性。
本文利用短波通信领域知识和深度学习中开集识别技术,提出了一个基于多模态特征融合的短波信号开集调制识别方法,旨在从复杂短波电磁环境中准确地识别接收信号的调制类型。所提方法利用接收信号的多模态特征表示作为深度网络输入,并通过不同模块实现了信号样本的数据增强、不同模态的特征融合以及深度特征学习,从而提高了短波信号开集调制识别性能。通过短波信道模型建立了短波调制信号数据集,并进行了各种对比实验和消融实验以验证所提方法的有效性和鲁棒性。实验结果表明:所提方法的性能明显优于现有的开集调制识别方法。在未来工作中,将通过实际采集的短波通信信号数据集进行实验,并且进一步研究如何通过有限样本来识别新的未知调制类型,从而提高该方法对实际短波调制识别场景的适应能力。
  • 国家自然科学基金(61971391)
参考文献 引证文献
排序方式:
[1]
WANG J, DING G, WANG H. HF communications: Past,present, and future[J]. China Communications, 2018, 15(9): 1-9.
[2]
GIESBRECHT J E, CLARKE R, ABBOTT D. Modulation recognition for HF signals[C]// Smart Structures, Devices, and Systems II. Bellingham: SPIE, 2005: 501-512.
[3]
PENG S, SUN S, YAO Y D. A survey of modulation classification using deep learning: Signal representation and data preprocessing[J]. IEEE Transactions on Neural Networks and Learning Systems, 2021, 33(12): 7020-7038.
[4]
ZHANG F, LUO C, XU J, et al. Deep learning based automatic modulation recognition: Models, datasets, and challenges[J]. Digit Signal Processing, 2022, 129: 103650.
[5]
DOBRE O A, ABDI A, BAR-NESS Y, et al. Survey of automatic modulation classification techniques: Classical approaches and new trends[J]. IET Communications,2007, 1(2): 137-156.
[6]
AZZOUZ E E, NANDI A K. Automatic identification of digital modulation types[J]. Signal Processing, 1995, 47(1): 55-69.
[7]
HAN L, GAO F, LI Z, et al. Low complexity automatic modulation classification based on order-statistics[J].IEEE Transactions on Wireless Communications, 2016,16(1): 400-411.
[8]
MOBASSERI B G. Digital modulation classification using constellation shape[J]. Signal Processing, 2000, 80(2): 251-277.
[9]
CHEN Z, WANG Z, XU D, et al. Learn to defend: Adversarial multi-distillation for automatic modulation recognition models[J]. IEEE Transactions on Information Forensics and Security, 2024, 19: 3690-3702.
[10]
HOU D, LI L, LIN W, et al. CIST: A convolutional transformer framework for automatic modulation recognition by knowledge distillation[J]. IEEE Transactions on Wireless Communications, 2024, 23(7): 8013-8028.
[11]
ZHANG J, WANG T, FENG Z, et al. Toward the automatic modulation classification with adaptive wavelet network[J]. IEEE Transactions on Cognitive Communications and Networking, 2023, 9(3): 549-563.
[12]
ZHANG X, LI T, GONG P, et al. Open set recognition of communication signal modulation based on deep learning[J]. IEEE Communications Letters, 2022, 26(7):1588-1592.
[13]
CHEN Y, XU X, QIN X. An open-set modulation recognition scheme with deep representation learning[J].IEEE Communications Letters, 2023, 27(3): 851-855.
[14]
CHEN Y, ZHU L, JIN C, et al. Boosting open-set RF signal recognition under low snr condition in feature space[J]. IEEE Communications Letters, 2024, 28(3):627-631.
[15]
LI T, WEN Z, LONG Y, et al. The importance of expert knowledge for automatic modulation open set recognition[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2023, 45(11): 13730-13748.
[16]
WATTERSON C C, JUROSHEK J R, BENSEMA W D. Experimental confirmation of an HF channel model[J].IEEE Transactions on Communication Technology,1970, 18(6): 792-803.
[17]
GENG C, HUANG S, CHEN S. Recent advances in open set recognition: A survey[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2020, 43(10): 3614-3631.
[18]
杨琳, 许小东, 路友荣, 等. 常见数字通信信号的谱线特征分析[J]. 电子与信息学报, 2009, 31(5): 1067-1071.
YANG Lin, XU Xiaodong, LU Yourong, et al. Spectrum line feature analysis for digital communication signals[J]. Journal of Electronics & Information Technology, 2009, 31(5): 1067-1071.
[19]
HE K, ZHANG X, REN S, et al. Deep residual learning for image recognition[C]// Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.New York: IEEE, 2016: 770-778.
[20]
SUN Y, CHENG C, ZHANG Y, et al. Circle loss: A unified perspective of pair similarity optimization[C]// Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. New York: IEEE, 2020:6398-6407.
[21]
ZHOU D W, YE H J, ZHAN D C. Learning placeholders for open-set recognition[C]// Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. New York: IEEE, 2021: 4401-4410.
[22]
VERMA V, LAMB A, BECKHAM C, et al. Manifold mixup: Better representations by interpolating hidden states[C]// International Conference on Machine Learning. PMLR. New York: Curran Associates, Inc. and International Machine Learning Society, 2019: 6438-6447.
[23]
International Telecommunication Union. Recommendation ITU-R. F. 1487: Testing of HF modems with bandwidths of up to about 12 kHz using ionospheric channel simulators[S]. Geneva: ITU, 2000.
[24]
田伟. 航空短波数据通信基带信号处理[M]. 北京: 清华大学出版社, 2022.
[25]
BENDALE A, BOULT T E. Towards open set deep networks[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. New York: IEEE,2016: 1563-1572.
2025年第46卷第6期
PDF下载
127
56
引用本文
BibTeX
文章信息
doi: 10.12347/j.ycyk.20250723001
  • 接收时间:2025-07-23
  • 首发时间:2026-03-13
补充材料
相关文章
文章信息
作者
出版历史
  • 收稿日期:2025-07-23
  • 修回日期:2025-07-29
基金
国家自然科学基金(61971391)
作者信息
    中国科学技术大学 合肥 230026
参考文献
分享链接
https://castjournals.cast.org.cn/joweb/ycyk/CN/10.12347/j.ycyk.20250723001
分享至
全文二维码

扫描看全文

引用本文
BibTeX
本文的引用情况
2种不同金属材料的力学参数

Family
属数
Number of
genus
种数
Number of
species
占总种数比例
Percentage of
total species (%)

Genus
种数
Number of
species
占总种数比例
Percentage of total
species (%)
鹅膏菌科Amanitaceae 2 11 5.26 鹅膏菌属 Amanita 10 4.78
小菇科 Mycenaceae 2 12 5.74 丝盖伞属 Inocybe 5 2.39
多孔菌科 Polyporaceae 8 14 6.70 蜡蘑属 Laccaria 5 2.39
红菇科 Russulaceae 3 23 11.00 小皮伞属 Marasmius 6 2.87
小菇属 Mycena 11 5.26
光柄菇属 Pluteus 5 2.39
红菇属 Russula 17 8.13
栓菌属 Trametes 5 2.39
关闭全屏